What Is Fair Value on a Stock Card and How Is It Estimated?
The midpoint of a reference range our model estimates from earnings, cash flow and peer valuation. It is an estimate — when confidence is low the error is wide.
Fair value is the central estimate of a reference price range that Stock Compass calculates for each company from its earnings, dividends, book value and analyst targets. It is not a prediction of where the price will go; it is a rule-based answer to "what would this share be worth if a few conventional valuation methods were applied to today's reported numbers?"
How it is calculated
The model runs up to four independent methods and then blends them:
- Earnings-based. Earnings per share (from P/E TTM) multiplied by a fair P/E: the sector's median P/E, adjusted for the company's ROE and revenue growth and capped. Skipped for financial companies.
- Dividend-based. The annual dividend per share divided by a target yield for the sector. Skipped for financials and non-payers.
- Book-based. Book value per share (from P/B) multiplied by a justified P/B derived from ROE, kept inside a sensible band. Skipped for financials with an ROE anomaly.
- Analyst consensus. The mean analyst target price, used as a fourth method.
The blend treats the analyst target as primary but guards against its usual optimism: if the in-house median is below the analyst target, fair value is the average of the two; if it is higher, the lower analyst target is used. An analyst-only estimate is discounted by 20% and flagged analyst_only_haircut; with no analyst target, the in-house median is used. The range runs from the lowest to the highest method value, widened by 5%. ETFs and leveraged products get no fair value.
Two derived outputs sit next to it. Margin of safety = (price minus fair value) divided by fair value, in percent — so negative means the price is below the estimate. Confidence is high when the in-house methods and the analyst target agree within 15%, medium up to 30%, low beyond that; analyst-only estimates are medium at best; an ROE anomaly forces low.
Rules of thumb
| Price vs fair value | Zone label |
|---|---|
| 15% or more below | Deep value |
| 5% to 15% below | Undervalued |
| Within 5% either way | Fairly valued |
| 5% to 20% above | Somewhat expensive |
| More than 20% above | Expensive |
The zone boundaries are the app's conventions; read them together with confidence: "deep value, low confidence" is a weaker statement than "fairly valued, high confidence".
What it does not tell you
- Timing. A stock can sit 30% below fair value for years. The model has no opinion about when, or whether, the gap closes.
- Cyclicals at the wrong moment. The methods use current earnings; at the top of a cycle they overstate value, at the bottom they understate it. The model raises a cyclical warning for these sectors.
- Truth when the inputs are distorted. Heavy buybacks shrink book equity and inflate ROE; the model flags this as an ROE anomaly, drops confidence to low, and the verdict says the analyst target is the more reliable side.
In Stock Compass
Fair value appears in the valuation row of the stock card in the full card view: the estimate, the signed margin of safety, a zone chip, and a low-confidence badge when applicable; hovering shows the range, confidence and warnings. In the card verdict it drives the valuation dimension: the summary states the distance from fair value and the zone, and a "cheap" rating is downgraded to neutral when confidence is low or an ROE anomaly is present. The verdict also checks fair value against the analyst target and explains which side to weigh more for that stock.
Fair value itself is not a rule indicator in the strategy builder, but its companion margin_of_safety_pct is (Fundamentals group, default margin_of_safety_pct < -10, where negative means below fair value). When valuation and price trend disagree, the guide Stop chasing the market explains why both can be right on different horizons.
FAQ
Why does fair value differ from the analyst target?
The analyst target is one of the model's inputs, not the whole answer. The in-house methods are deliberately conservative, and the blend leans toward the lower of the two when they disagree. A large gap is itself information: there is little consensus on what the stock is worth.
What does low confidence mean?
The methods disagreed with each other by more than 30%, or the estimate rests on one method, or an ROE anomaly makes the book-based inputs unreliable. Treat the fair value as a rough reference and lean on the range rather than the single number.
Why is there no fair value for my ETF or for some HK stocks?
ETFs and leveraged products are excluded because company valuation methods do not apply to a basket. For some HK and A-share listings the fundamentals feed is incomplete, so too few methods can run and the model returns nothing rather than guess.
Is the margin of safety sign the usual way round?
No — be careful. The app's margin_of_safety_pct is (price minus fair value) over fair value, so a negative number means the price is below the estimate. Writing > 20 in a rule selects stocks 20% above fair value, not stocks with a 20% margin.
Definition only — not investment advice.